As the CCF-GAIR conference robots and drones, the finalist of the “Commercial Scenes of Robotics†module, distinguished guest of the Changjiang Scholar of the Ministry of Education, Chairman of IEEE Beijing Robotics and Automation, Honorary Director of Beihang Robot Research Institute, advanced Professor Wang Tianmiao, the chief expert in the field of robotics, brought a keynote speech titled "Thinking of Robot Development - Academia to Industry" for the conference.
Professor Wang Tianmiao first analyzed the current macro environment from a macro perspective. He pointed out that the current domestic and international macroeconomic situation has gradually improved, coupled with the unique market environment in China, application of indexing and multidisciplinary cross-fertilization in recent years, making the current artificial intelligence and robotics industry has actually become A kind of long-term recessive just needs, in which the intelligent robot technology of perception, decision-making, action and other categories has become a strategic high-tech industry in various developed countries in the world.
Subsequently, Professor Wang Tianmiao briefly reviewed the development of artificial intelligence and robots from four stages.
1. A manufacturing robot represented by smart mechanical components.
2. Special robots represented by environmental adaptation and simple sensing.
3. Service robot represented by human-computer interaction and flexible structure.
4. Intelligent robot represented by intelligent and bionic technologies.
Next, Professor Wang Tianmiao shared a number of hot issues in the field of artificial intelligence and robotics in one breath, and clearly pointed out that it is currently in the bonus period where robots and artificial intelligence flourish. The hot topics shared by Prof. Wang include: coordination and assistance robots for customer service, shopping guide, medical and chat; home smart butler or caring nanny robots; driverless technology; VR/AR technology; smart health robots; rigid-soft coupling software Structure and artificial perception of electronic skin; micro-small special-function robots and so on.
At the same time, Professor Wang Tianmiao also concluded that the key technologies in the field of artificial intelligence are: deep learning and cognitive evolution, acquisition and processing of big data, artificial intelligence chips, semantic understanding, conceptual expression and knowledge reasoning.
Professor Wang Tianmiao believes that in the days when artificial intelligence and robots are just in a prosperous rising period, technological changes and market conditions can be described as a result. How to transform scientific research achievements of academic circles into successful products in the industry in the current complex and ever-changing ecological environment? It is the key and difficult point that the industry really has to face and solve.
Prof. Wang Tianmiao made it clear that only the strength, talent, and industrial capital of the country can be successfully promoted by the mutual support and coordination of these three parties. However, in the current environment, there are two key obstacles, which are manifested in the following two aspects.
1. In nature, there is a completely different assessment orientation among colleges and universities across the country. At the same time, various legal disputes encountered during the transformation of technological achievements from the academic world to the industrial sector are likely to cause damage to schools. In addition, how to buy IP, how much of the shares should be, who will assess, and how to improve efficiency, all these issues restrict the "nature" in the transformation process.
2. In terms of positioning, mentors or academics are often presented with cutting-edge papers, and companies often require patents and monographs. They are responsible to customers and shareholders. They must be practical, convenient, and often have a long process of transformation. The positioning of the two exists. Obvious inequalities and contradictions. In addition, if it is a scholar, to start a business from this perspective, it will be very difficult. At this time, if the students are allowed to start businesses after graduating in the form of consultants, the success rate is often high. This is an important role of focus and is also a problem of “positioningâ€.
At the end of the speech, Professor Wang Tianmiao stated that innovation and entrepreneurship is a long-term process, which requires the combination of technology and capital, the need for down-to-earth partners, and the need for synergies between policy and the environment. It is very complicated. However, he firmly believes that this is the era of innovation and entrepreneurship in the field of robotics and artificial intelligence. He warmly welcomes people of insight to bravely embrace this spring.
The following is a record of the scene:
Thank you for the introduction of Teacher Bi. I am very happy to share with you 25 minutes of my thoughts on the academic world and the industry. I mainly talk about two parts. One part wants to talk about hot spots. Second, I talk about the difficulties from the academic world to the industry. Confused, and some of your own thoughts.
In the whole development process today, I personally think that in the future in the field of artificial intelligence and robots in the medical field and smart manufacturing, as well as in the medical, sports, and chat areas, this is a very important outlet. why? There are two main reasons for this. First, the cross-convergence of new technologies such as robotic artificial intelligence, new energy, and the Internet of Things has reached a critical point. This aspect may promote the development of the entire society, including changes in the industrial structure. Second, the Chinese market is very large, and China's manufacturing has its unique advantages, coupled with our country's strong support in this regard.
In this context, artificial intelligence and robots are actually rigid demands for a long time. The reason is that it is necessary for our country to manufacture, equip, and provide basic equipment, and it plays an important role in national economic security. Looking at the entire scope of the current situation, a lot of normal irreversibility has promoted the development of artificial intelligence and robotics, and what is irreversible, such as rising wages, recruitment difficulties, aging, and increased medical expenses, etc., which are irreversible. At the same time, Wisdom life, turn over and stimulate and release new needs, including safety, freedom, health, happiness, and even include some adventures, including some innovation. This staggered, it makes people feel that it is a rigid demand for quite some time. Because of this, in the United States, Japan, including Germany, smart technology, or intelligent robot technology, is considered as a national strategic plan. They believe that perception, decision-making and action will promote many aspects of development.
Regarding smart hotspots, we typically call it the Gartner 2016 smart machine, which is basically starting to be profitable, such as cloud recognition, like our virtual reality, and software tools based on knowledge development. The crest is the current investor, and what people dream about, some of which are promoted, including what we call autonomous vehicles, perception, machine learning, and others. Initially it was a section of innovation and climbing that included smart robots, commercial drones, and some new machine assistants. In this context, we look at these hot spots and it is very clear. Let's share with you the history of the development of robots, which has always been accompanied by the development of artificial intelligence technology.
As early as 1945, the United States had to engage in atomic bombs, conduct mechanical and geared operation, and invented the operation of robots and DH parameters in the following 54 years. Sifu broke through the control of the machine tool. When I really got into the market, I found that the imagination was not so big. In the future, it was industrialized by Japan. The legendary company first used it in replicable cars and motorcycles. There are two things in it. Artificial intelligence, such as visually autonomous assembly, at the same time, Stanford University engaged in a mobile robot, can be on the rostrum, but basically failed. That failure was also a result of the low tide of artificial intelligence. It was followed by another era. In the late 1980s and early 1990s, a new US high-tech plan emerged. At that time, the smart system of the CMU was the prototype of the smart car. At that time, the space and the underwater were promoted. From the resource node at that time, the space at that time was small and it was hoped that the service could be entered.
There are several things happening here, IRobot is listed, there are encyclopedias, and it was also the United States. There have been no major breakthroughs so far, not where people's pain points are, and if there are too many pain points, washing the floor, washing dishes, etc., it is found that the cost is too high and there is another one, pushing the cooperating robot forward and there is also an alpha Dogs, who have proved that memory and case studies are better than people, once they break through, they find that many decision-making things can be added. Google acquired eight cluster companies. From this point of view, will there be new prospects.
From the coordinate point of view, we often put the required function, the ordinate is the complexity, the results of the reactor can only be carried out in the laboratory for some experiments, because of reliability, cost, including multiple systems. Nowadays, based on Internet technology, it is dismantled. For example, deep learning or recognition is placed on the cloud, sensors are placed in space or at home, and application-driven, and the robot concept is pushed to the front line.
In this context, the development of the robot was formed. We know that the economy of the entire world is slowly climbing, that is, 2%-3%. However, the growth rate of industrial clusters is 12%, which is still global, and China's growth rate is 17%. With 240,000 newly installed in the world, China is 6.7. From this perspective, Asia is still the most important area in China, followed by Europe and the Americas. What this film wants to share with you is to have confidence in China's innovation and entrepreneurship, or in a national enterprise. Looking at it now, it is often said that robots are all four big families. This is actually the case. However, they are currently concentrated in automobiles and motorcycles, including some other electronic equipment. The current occupancy is about 90%. Think about it. Three years ago, foreign countries were 96%, 92%, and 88%. At present, this film reflects that the growth rate of domestic cluster brands is about 33%. From this perspective, it will not be long before the mid- to low-end will almost break through 60%, and the high-end will have a process because it involves parts and components and involves brands and solutions.
In the future, clusters will be the hotspot. Coordination and mutual assistance are very important directions of development. At present, the growth rate of special robots is about 11%, and that of home robots is about 28%. It is predicted by the International Robot Federation that it is about 20 billion US dollars, probably in 2018. This number and industry ratio, the industry's global scale last year was about 30 billion US dollars, which means that the space for service robots in the next 5-10 years may be very large. What are the reflections, such as the percentage of defense, medical care, and services, and the proportion of personal robots in home entertainment.
The current difficulties, whether abroad or domestic, want to break through such a breakthrough. For example, we often understand Amazon, and we want to develop from the interaction into the entrance. Google also announced table lamp interactions in the near future. The reason is still because there is a balance between pain points and costs, as well as the reliability of the implementation technology. This picture is given by GF Securities and is very interesting. Demand, technology, cost, weak demand and rigid demand, in this coordinate system, there are high quality, low product, help to tap the pain points of the robot demand.
Judging from the current trend, robots centered on customer service, shopping guides, medical care, and chat are currently a hot spot in this five-year period. Then there will be what we call unmanned, as well as the just-mentioned stewards or nannies. This process is very high. This process is modeled and demonstrated.
Share a few hot spots with you. In a broad sense, artificial intelligence is actually cognitive learning, decision making and thinking. You can understand it as the brain of a physical robot. Perception, we think is a human-computer interaction and feelings. RT, we think it is a movement and operation. On these issues, the current hotspot, we believe that it should be in the direction of software software and micro-small development of the ontology, but also the cutting-edge and bio-manufacturing of intelligent hardware.
Toyota bought new products from Google, including artificially climbing under the table, a lot of hydraulic CNCs, and some drives. There is also wearable, micro- and nano-made, including testing in a narrow space. Da Vinci's tools are a contribution in this regard. The rigid-flexible coupled software domain artificially perceives the electronic skin, and the director of the MIT in the morning has already reported that the top scientific reports in the world in 2012 and 2015, mainly focused on MIT and universities, there are some drivers that can use silicone, 3D printing is complete, imagine that if this breakthrough, may be a reshuffle of the robot, including just the yellow teacher to talk about his dream. The birth of such a soft robot may be very close to practicality. Through our experiments, we also found a feature whose cost is 1/5 of that of traditional hardware robots.
In the area of ​​artificial intelligence robots, the main focus is on cognition and deep learning, especially on evolutionary learning. The content is continuously updated and the corresponding coefficients are adjusted. Now individual learning is still relatively good, but it is still relatively big challenge to concentrate, so that voice and visual can be applied as soon as possible, and the other is an artificial chip. These points, I think, are all very important directions. It is expected that there will be voice recognition. This is mainly in games, movies, and sports. It involves natural language understanding, including somatosensory.
Industrial robots can not be programmed automatically through actions. If so, the robot's progress can be greatly reduced. Including what we call the AR technology breakthrough, this will free up your hands and match the background knowledge with what you see through the eyes, thus reminding you of many things. This shows an important direction. With so many Internet of Things, there must always be an agent. Naturally, many people want to make breakthroughs in robots at home or in the community. Here, I hope everyone in this area must think about what is irreplaceable. If not, everyone will do this. This tune is easy to get smaller. In terms of operations, learning, policy materials, multi-machine and man-machine coordination, in the hot areas of application, mainly in the structure of the impersonation, machine assistants, to accompany each other, and then there is a special drone system, unmanned vehicles, Unmanned boats, drones, including medical machinery. Sometimes I think that time is slowly changing many of our things. Actually, we don't know every single day and every month. Looking back for 5-10 years, we find that the world's shape is changing, the industrial structure is changing, and the way of life is changing. In such changes, how to turn scientific research and innovation into industry is particularly important.
Fortunately, in the past 10 years, I have incubated with many colleagues and participated in the guidance of a number of representative companies, including Ninebot, who was just introduced on the stage by Zhang. This is a group of two Beihang students. One is Wang Ye and Lu Feng. For example, Segway used to It is 2B. They boldly marched towards 2C. On January 8th, 2016, intel gave them a platform, and the whole process got millet, redwood, and even other investments. For nearly 20 years, medical robots have cooperated with the Navy's General Hospital. From the doctor's point of view, whether they can have 10%-20% of the operations without opening the skull, through the rice, put the medicines in, this benefit can reduce the cost of medical treatment, and The patient may be discharged from the hospital within three to four days. If it is a craniotomy, it will be three months or even longer. Cost, technology will be reduced, such technology will get rid of the traditional framework, surgical planning, robots, including some safe control. After this aspect, we got the support of Huaxia Happiness, Puhua and others. President Xi also visited the cooperation with Jishuitan in June this year. Google used a robot to avoid X-rays. It used to be more than 50 times and it has now avoided it. In this regard, the radiation to doctors has been greatly reduced. The doctors like it very much, and the precision has been increased from 2 mm to 0.5. There are currently more than 2,000 cases. And successfully listed on the three boards. This is after my students graduated to the Academy of Sciences, and left from the Academy of Sciences to use the operating system to change the structure of the current car, called smart meters, and was recruited as a LEES developer in June this year. They are large companies, Toyota, BMW. Looking at cars with LEES, there are more than 100 CPUs in a car and more than 200 sensors. In this environment, software-defined car functions become a must.
And so on, it also includes dexterous hands, big data. There are some reflections from such an innovation to the transformation process of the industry that we are talking about. I think the significance of innovation and entrepreneurship should actually be deeply understood. That is, apart from exploring the unknown, it is the contribution to the economy. On this issue, in fact, society, schools, and the country not only hope to pay taxes for employment, but also hope to become great companies, and to give China a glory in international competitiveness. After learning these students will feedback to the community, and even feedback to his alma mater.
On this issue, I believe that the guiding ideology is very important, and we must understand that 20% of basic research can be separated from the market, free from any utilitarianism, and should be free to explore and think. But the vast majority of technology research and development, its value should be to the contribution to the economy. There are three links in this guiding ideology: national strength, talent, and industrial capital. The intersection of the three can promote its transformation. The reason is very simple, but it is very difficult in the process of doing it. Because from the academic world to the industrial transformation, what the industry understands is to hope that you have a bit of pain and maturity, and also hopes that this team will concentrate on monographs and hope that you should look like Market learning should learn to operate. But these three things hope to be like this, but it is very difficult to do it. Why is it difficult? First of all, the nature of the problem, from the current point of view of the details, the country within the school, due to the different orientation of the assessment, it is difficult to complete. Because in this process, we have to go quite a long way. And in the entire process of legal disputes, every company in the world has it, and it will not harm the school. In addition, how to buy IP, how much of the shares should be, who will assess, how to improve efficiency, these issues restrict the nature of the transformation process, as well as positioning. Mentors or academics are often new, often based on papers, and companies often have monographs, take responsibility for customers, be practical and convenient, and have a long-term transformation process. On this issue, I think this problem will arise.
I would like to take a simple case and say that the case of Leonardo da Vinci, for example, was invented by Stanford in 1990. It was not until 2015 that everyone looked at it. It took 15 years to get this kind of proof. Actually, this The side reflects that this process is necessary in hard technology or in the high end of the robot. Here is my history of combing Da Vinci's venture. There is capital, there is a way to change the way doctors operate, and there is high-definition. Once at an international conference, he asked China that this robot should be a doctor who had a problem. He asked Europe. Europe is a robot and it is a device. Asked the French, the French said they should be looking for an insurance company. In thinking about the problems solved here, the country is now promoting this, allowing more than 50% of inventors and creators to have it, and some of them are already 70% scrutinized. Can they be simplistic and accelerate? Meeting. Including the University of Pennsylvania yesterday, it was basically 5% or 10%, which was necessary for school development. For example, the IP problem, if each IP is 100,000 or 200,000, will find this speed will accelerate, the national laws and regulations are very important in this regard.
Finally, I want to talk about positioning. If you are a scholar, it is very difficult to start a business from this perspective. Otherwise, from the consultant's point of view, let your students graduate from starting business after graduation. This success rate is very great. This is because there is only hope and concentration. These ideas came from Antarctica. Experiments were conducted in Antarctica. In the harsh environment, such experiments were conducted in half a month. The use of wind energy and solar energy to charge the battery and enable it to operate using renewable energy is also the time to think about the academic community. The idea of ​​industrial transformation.
On this road, every entrepreneur knows that this road is very long and involves all aspects. It is just an atom, that is, what we call the combination of innovative technology and capital. The real development must be partnership, but also policy, but also the environment, but also adults to teach him. In any case, I think that in today's world, the era of innovation and entrepreneurship has arrived. Everyone must ask themselves if they have such a passion. Is there any irreplaceability? Is it possible to achieve it? Tolerance should be more patience, not quick success. I believe that each of us can find our own position in the promotion of the national wave of innovation and entrepreneurship. Thank you.
The host Bi Yalei (Secretary-General of the Shenzhen Computer Association): I think Teacher Wang today gave us a concluding speech. Teacher Wang was the earliest professor involved in the industrialization of the robotics industry. As he said today, there are many Thinking. A small problem, among the students in your world, you can give a student startup project that is most promising to you. You just talked about doing the conversion. Actually, I also do transformation in the organization and often help the teacher, Students start a business, can you tell a case?
Wang Tianmiao: In the moment, I think that Ninebot is very good, there is also a medical robot Remebot, there is a Segway Robot, these have several characteristics, the first is a high threshold; second, in the domestic or worldwide Inside the same starting line. What remains is two things. Their keenness of the market and the team's personality pattern determine how far he can go.